I'm using postgreSQL and mapping a simple database table. The primary ID of my table is a sequence. When adding rows to the table manually, I never specify the ID - it is autogenerated by the database because the sequence is specified as the default value.

After reading the Hibernate manual, I am unsure how to model this behavior in the mapping. The closest I can see is something like this:

<id name="id" type="int" column="id" unsaved-value="0">
<generator class="sequence"/>
<param name="sequence">mmseq</param>
</generator>
</id>

but the docs say using this method requires two SQL statements to insert a new row into the database: one to get the sequence value, and the other to do the insert. That seems wasteful to me. I've never used two statements with plain old JDBC - just don't specify the ID and the db generates it.

Can hibernate do that?

How did you get the id for your object without querying for it ?

Hmmm, your question makes me realize my head isn't completely wrapped around the Hibernate paradigm yet.

I was thinking along the lines that all I want to do is add a new row to the database and be done with it. In this particular case, I probably won't refer back to the row I just added for quite awhile. If I was doing this the old JDBC way, it would be a simple SQL insert statment and I would not specify a value for the ID column.

But I'm now guessing Hibernate can't do that because it wants to associated my persistant java bean with the row I just added to the DB. And to do that, it needs to set the bean with the ID of row just added. Doing that with a sequence requires two queries.

Do I have any of that right?

Exactly, for example after saving the object you could change some of its properties, and hibernate must know which row to update ...